Research map / Image generation / Fast diffusion sampling
TrajFlow: Nation-wide Pseudo GPS Trajectory Generation with Flow Matching Models
Peiran Li, Jiawei Wang, Haoran Zhang, Xiaodan Shi, Noboru Koshizuka, Chihiro Shimizu, Renhe Jiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use flow matching instead of DDPM denoising (300 steps) (ablation) | TrajFlow · Japan nationwide | DTW median (km): 26.125 -> 10.977 | Improved |
| add OD prediction and reconstruction (ablation) | TrajFlow · Japan nationwide | DTW median (km): 100.271 -> 10.977 | Improved |
| add RDP harmonization (ablation) | TrajFlow · Japan nationwide | DTW median (km): 24.690 -> 10.977 | Improved |
| add RDP and OD at urban scale (ablation) | TrajFlow · Central Tokyo | DTW median (km): 8.179 -> 20.350 | Hurt |
| add RDP and OD at metropolitan scale (ablation) | TrajFlow · Tokyo Metropolis | DTW median (km): 14.416 -> 18.167 | Hurt |
| add OD prediction on Chengdu open data (ablation) | TrajFlow · Chengdu | Density JS: 0.0140 -> 0.211 | Hurt |
| use TrajFlow instead of DiffTraj | DiffTraj · Japan nationwide | DTW median (km): 451.042 -> 10.977 | Improved |
| increase DDPM sampling steps from 10 to 300 (ablation) | TrajFlow w/o FM · Japan nationwide | DTW median (km): 495.961 -> 26.125 | Improved |
| use RDP instead of DCT for trajectory compression (ablation) | trajectory compression · 5 params | DTW (km): 3.23 -> 2.77 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DiffTraj — DiffTraj: Generating GPS Trajectory with Diffusion Probabilistic Model (NeurIPS 2023) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Satellites Reveal Mobility: A Commuting Origin-destination Flow Generator for Global Cities (NeurIPS 2025 · similarity 0.86)
- TrajMamba: An Efficient and Semantic-rich Vehicle Trajectory Pre-training Model (NeurIPS 2025 · similarity 0.86)
- GTG: Generalizable Trajectory Generation Model for Urban Mobility (AAAI 2025 · similarity 0.86)
- UniTraj: Learning a Universal Trajectory Foundation Model from Billion-Scale Worldwide Traces (NeurIPS 2025 · similarity 0.86)
- PLMTrajRec: A Scalable and Generalizable Trajectory Recovery Method with Pre-trained Language Models (NeurIPS 2025 · similarity 0.86)
- SeoulMMOD: A Large-Scale Multimodal Origin-Destination Flow Benchmark (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
